U.S. patent application number 14/280550 was filed with the patent office on 2014-09-11 for method and system for tank refilling.
This patent application is currently assigned to Honda Motor Co., Ltd.. The applicant listed for this patent is Honda Motor Co., Ltd.. Invention is credited to Ryan HARTY, Steve MATHISON.
Application Number | 20140251495 14/280550 |
Document ID | / |
Family ID | 44814769 |
Filed Date | 2014-09-11 |
United States Patent
Application |
20140251495 |
Kind Code |
A1 |
HARTY; Ryan ; et
al. |
September 11, 2014 |
METHOD AND SYSTEM FOR TANK REFILLING
Abstract
Disclosed is a simple, analytical method that can be utilized by
hydrogen filling stations for directly and accurately calculating
the end-of-fill temperature in a hydrogen tank that, in turn,
allows for improvements in the fill quantity while tending to
reduce refueling time. The calculations involve calculation of a
composite heat capacity value, MC, from a set of thermodynamic
parameters drawn from both the tank system receiving the gas and
the station supplying the gas. These thermodynamic parameters are
utilized in a series of simple analytical equations to define a
multi-step process by which target fill times, final temperatures
and final pressures can be determined. The parameters can be
communicated to the station directly from the vehicle or retrieved
from a database accessible by the station. Because the method is
based on direct measurements of actual thermodynamic conditions and
quantified thermodynamic behavior, significantly improved tank
filling results can be achieved.
Inventors: |
HARTY; Ryan; (Long Beach,
CA) ; MATHISON; Steve; (Torrance, CA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Honda Motor Co., Ltd. |
Tokyo |
|
JP |
|
|
Assignee: |
Honda Motor Co., Ltd.
Tokyo
JP
|
Family ID: |
44814769 |
Appl. No.: |
14/280550 |
Filed: |
May 16, 2014 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
12982966 |
Dec 31, 2010 |
8783303 |
|
|
14280550 |
|
|
|
|
Current U.S.
Class: |
141/4 |
Current CPC
Class: |
F17C 2223/036 20130101;
F17C 2250/0631 20130101; F17C 2201/0109 20130101; F17C 2201/054
20130101; F17C 2221/033 20130101; F17C 2270/0139 20130101; F17C
5/06 20130101; F17C 2203/0619 20130101; F17C 2223/0123 20130101;
F17C 2223/0161 20130101; F17C 2223/0153 20130101; F17C 2225/036
20130101; F17C 2203/0604 20130101; F17C 2250/0694 20130101; F17C
2270/0168 20130101; F17C 13/026 20130101; F17C 13/025 20130101;
F17C 13/023 20130101; F17C 2260/022 20130101; F17C 2203/0624
20130101; F17C 2221/012 20130101; F17C 2260/026 20130101; F17C
2250/043 20130101; F17C 2250/0491 20130101; F17C 2260/023 20130101;
F17C 5/007 20130101; F17C 2265/065 20130101; Y02E 60/321 20130101;
F17C 2250/0439 20130101; Y02E 60/32 20130101; F17C 2270/0176
20130101; F17C 2260/025 20130101; F17C 2250/0495 20130101; F17C
2225/0123 20130101; F17C 2250/0443 20130101 |
Class at
Publication: |
141/4 |
International
Class: |
F17C 5/06 20060101
F17C005/06; F17C 13/02 20060101 F17C013/02 |
Claims
1-11. (canceled)
12. A method of operating a hydrogen gas filling station
comprising: obtaining a first set of parametric data corresponding
to a hydrogen powered vehicle; obtaining a second set of parametric
data corresponding to station capabilities; obtaining a third set
of parametric data corresponding to a refueling ambient;
calculating a MC value based on the parametric data obtained; and
determining a projected fill time that is predicted to produce a
gas final temperature T.sub.final no greater than a target
temperature T and achieve a state of charge of 100% within the tank
according to Equation T Final = m cv C v T Adiabetic + M C T
Initial ( M C + m cv C v ) ; [ B ] ##EQU00015## and determining a
target pressure P.sub.target that is predicted to produce a state
of charge of 100% within the tank.
13. The method of operating a hydrogen gas filling station
according to claim 12, wherein: the first set of parametric data is
obtained directly from the hydrogen powered vehicle using a
communication protocol selected from a group consisting of RFID,
HVAS and IRDA.
14. The method of operating a hydrogen gas filling station
according to claim 12, wherein: the first set of parametric data is
obtained by identifying the hydrogen powered vehicle and accessing
a parametric database maintained outside the vehicle.
15. The method of operating a hydrogen gas filling station
according to claim 12, wherein: the first set of parametric data is
assigned default values and determining an adjusted final target
density sufficient to achieve the P.sub.target at the
T.sub.final.
16. The method of operating a hydrogen gas filling station
according to claim 12, wherein: the first set of parametric data
includes at least one parameter selected from a group consisting of
nominal working pressure, tank volume, refueling history, time
stamp of last refueling, trip odometer value, initial hydrogen
mass, initial gas temperature, maximum hot soak temperature,
maximum cold soak temperature, maximum defueling rate and maximum
fueling rate.
17. The method of operating a hydrogen gas filling station
according to claim 12, further comprising: maintaining first and
second hydrogen fill assemblies, the first hydrogen fill assembly
operating at a first pre-cooling temperature and the second
hydrogen assembly operating at a second pre-cooling temperature,
the first and second pre-cooling temperatures not being equal;
analyzing the first set of parametric data; and directing the
hydrogen powered vehicle to the hydrogen fill assembly that can
more efficiently provide a state of charge of 100%.
18. The method of operating a hydrogen gas filling station
according to claim 12, further comprising: calculating a target
pre-cooling temperature sufficient to achieve a 3-minute fueling
time using parametric information selected from a data set
consisting of vehicle fueling history, tank volume, lowest
historical initial state-of-charge percent, lowest historical tank
MC value, and ambient temperature; setting and maintaining a
hydrogen fill assembly at the target pre-cooling temperature for
the duration of a vehicle refueling operation.
19. The method of operating a hydrogen gas filling station
according to claim 12, further comprising: predicting a defueled
system temperature for a combined system heat mass; setting
T.sub.FinalDefuelCold; and initiating refueling using
T.sub.FinalDefuelCold.
20. The method of operating a hydrogen gas fill station according
to claim 12, further comprising: obtaining a first subset of the
first set of parametric data, the first subset corresponding to
parametric data associated with a first hydrogen tank onboard the
hydrogen powered vehicle; obtaining a second subset of the first
set of parametric data, the second subset corresponding to
parametric data associated with a second hydrogen tank onboard the
hydrogen powered vehicle; calculating T.sub.final1 using using the
first subset of parametric data and T.sub.final2 using the second
subset of parametric data; using the greater of T.sub.final1 and
T.sub.final2 to determine a target fueling speed; and using the
lesser of T.sub.final1 and T.sub.final2 determine P.sub.target.
21. The method of operating a hydrogen gas fill station according
to claim 12, further comprising: initiating a hydrogen gas fill;
adjusting the second set of parametric data during the hydrogen gas
fill to reflect measured station performance; adjusting the MC
value during the hydrogen gas fill to reflect variations in the
first set of parametric data and measured fill performance data;
calculating an adjusted fill time and P.sub.target to reflect
adjustments in the second set of parametric data and MC value; and
using the adjusted fill time to control the fill speed and end of
fill conditions.
22. The method of operating a hydrogen gas fill station according
to claim 17, wherein: the more efficient hydrogen fill assembly
provides a benefit selected from a group consisting of reduced
fueling time, higher state-of-charge percent, increased pre-cooling
temperature, reduced energy consumption and combinations
thereof.
23. The method of operating a hydrogen gas filling station
according to claim 12, the step of obtaining a first set of
parametric data corresponding to a hydrogen powered vehicle further
comprising: retrieving refueling time stamp data for the hydrogen
powered vehicle.
24. The method of operating a hydrogen gas filling station
according to claim 23, wherein: the refueling time stamp data is
retrieved from a HVAS RFID tag associated with the hydrogen powered
vehicle.
25. The method of operating a hydrogen gas filling station
according to claim 23, wherein: determining an elapsed time from
the refueling time stamp data; and adjusting fueling speed to allow
for a higher refueling speed for a longer elapsed time.
Description
PRIORITY STATEMENT
[0001] This application claims priority pursuant to 35 U.S.C.
.sctn.119(e) from U.S. Provisional Patent Application Nos.
61/326,375 and 61/332,919, the contents of which are incorporated
by reference, in their entirety.
BACKGROUND OF THE INVENTION
[0002] The safety and convenience of hydrogen tank refueling are
recognized as important considerations in determining the ultimate
success of hydrogen fueled vehicles in the marketplace. Under
current safety guidelines, the refueling of compressed hydrogen
tanks are to be conducted in a manner that prevents the tank from
overheating (temperatures exceeding 85.degree. C.) during refueling
and/or from overfilling the tank to a point at which the pressure
could exceed 125% of normal working pressure (NWP) at any time.
Because of the number of unknown parameters associated with
conventional hydrogen tank refueling procedures, the refueling
operations tend to be somewhat conservative, thereby trading
performance and efficiency, particularly with respect to end of
fill density (SOC %) and/or unnecessary levels of pre-cooling, for
an increased safety margin. A SOC of 100%, for example, corresponds
to a tank at NWP and 15.degree. C.
[0003] This tradeoff is especially significant in non-communication
fueling operations in which the parametric assumptions are even
more conservative. Because the hydrogen station does not have
information about the tank that it is filling, very conservative
assumptions need to be made for the system in order to encompass
the range of possible tank configurations and initial tank
conditions to avoid exceeding the system safety limits. In SAE TIR
J2601, the disclosure of which is incorporated herein by reference,
in its entirety, these conservative assumptions are incorporated
into a series of lookup tables for hydrogen tank filling. Working
from parameters including the tank volume, starting pressure,
ambient temperature and station pre-cooling set point, the lookup
tables are then used for determining a pressure ramp rate and final
target pressure. While application of these lookup tables tends to
provide for safe refilling under virtually all conditions and for
virtually all tank systems, given the conservative nature of the
associated assumptions, the resulting hydrogen tank filling
operation may take longer, achieve lower final fill pressures
and/or require lower hydrogen station pre-cooling temperatures than
necessary to fill a particular tank system.
[0004] An additional limitation of the refilling procedures defined
by SAE TIR J2601 is the lack of any method or procedure for a
hydrogen tank filling station to compensate or adjust for
situations in which its actual operating conditions fall outside of
the allowed tolerances. For example, if the pre-cooling temperature
is above the design set point as the result of multiple consecutive
refills, the lookup tables defined in SAE TIR J2601 cannot be used.
Efforts to avoid this out of specification condition can lead to an
overdesigned hydrogen tank filling station (excessive cooling for
ensuring that the pre-cooling target temperature is maintained),
thereby driving up station cost.
[0005] Conversely, failing to ensure that the pre-cooling target
temperature is maintained can inconvenience customers that are
unable to refill their tanks in a timely manner (as a result of
delays waiting for the pre-cooling temperature to come into
specification), thereby reducing customer satisfaction, station
revenue and/or repeat business. Further, operating a station with a
constant pre-cooling temperature regardless of current ambient
conditions results in excessive energy usage and reduced
well-to-wheel energy efficiency. In order to reduce energy use, a
hydrogen tank filling station should be operated at the highest
possible pre-cooling temperature that provides both
customer-acceptable refueling times and a satisfactory safety
margin.
BRIEF SUMMARY
[0006] The MC Method as detailed infra provides a new tank filling
model based on the total heat capacity of the hydrogen fueling
system and an advanced algorithm based on that model for improving
the performance of hydrogen filling stations under a broad range of
operating conditions. This algorithm, as applied stepwise in the MC
Method, can be used to enhance fueling performance pursuant to SAE
TIR J2601 through the use of additional thermodynamic information
about the tank system. The MC Method works at virtually any tank
system Normal Working Pressure (NWP) and with any compressed
hydrogen tank system, and it allows fueling at conditions that
operate outside of current SAE TIR J2601 tables, such as new
stations with -10.degree. C. pre-cooling, or previously existing
stations without the pre-cooling capability as specified in SAE TIR
J2601. Utilizing the MC Method will permit hydrogen filling
stations to improve their fill speed and fill quality (SOC %),
while enabling lower cost hydrogen stations to meet those
needs.
BRIEF DESCRIPTION OF THE DRAWINGS
[0007] Example embodiments described below will be more clearly
understood when the detailed description is considered in
conjunction with the accompanying drawings, in which:
[0008] FIG. 1 illustrates the modeling of a hydrogen storage tank
during refueling as an open system with an unsteady flow control
volume model. For the purpose of this disclosure, the control
volume is defined as the boundary between the gas and the liner
with heat being transferred through the boundary of the control
volume and into the liner of the tank.
[0009] FIG. 2 illustrates a temperature v. time curve for a
hydrogen tank refueling procedure, reflecting use of the adiabatic
temperature in calculating the heat transfer with the heat
transferred from the hydrogen being described by Equation [5]
infra.
[0010] FIG. 3 illustrates a constant heat flux model showing
temperature distribution dependent on time with adiabatic boundary
condition with a conservative assumption of no heat transfer from
the outside of the tank so that the actual final temperature in the
tank is expected to be slightly lower than the value calculated in
light of this assumption.
[0011] FIG. 4 illustrates a temperature distribution of a section
of a composite tank immediately after a vehicle refueling.
[0012] FIG. 5 illustrates a simplified representation of a hydrogen
tank having an imaginary characteristic volume of combined mass and
specific heat capacity MC, with adiabatic external boundary.
[0013] FIG. 6 illustrates a typical vehicle fill in 3 minutes with
a Type 3 tank that produces an end-of-fill MC value of 62 that then
tends to increase as the tank cools. This trend of MC v. time is
characteristic for a given tank system, and can be used in
predicting the temperature result of longer vehicle fueling
times.
[0014] FIG. 7 illustrates MC v. fill time for a "default" SAE TIR
J2601-70 MPa Type 4 tank. This graph is derived from the Type B
(7-10 kg) 70 MPa station tables of SAE TIR J2601.
[0015] FIG. 8 illustrates a potential Test Matrix for tank
characterization
[0016] FIGS. 9A and 9B illustrate MC v. U.sub.adiabatic/U.sub.init
for fills of a Type 3 tank from which the coefficients A and C can
be determined for both longer fill times, using a linear
approximation, and shorter fill times, using a logarithmic
approximation.
[0017] FIG. 10 illustrates .DELTA.MC v. time for fill times having
a duration of more than 3 minutes from which the coefficients g, k,
and j can be determined for describing the behavior of MC for fill
times in excess of 3 minutes.
[0018] FIG. 11 illustrates a comparison of the hydrogen station
operating envelopes to the existing refueling standards, showing
several gaps in the coverage of existing or anticipated operating
regimes.
[0019] FIG. 12 illustrates information required for fully utilizing
the MC Method for determining a fueling protocol under a given set
of conditions.
[0020] FIG. 13 illustrates the MC Method first step--determining
the fueling time based on a higher than ambient temperature soak
condition.
[0021] FIG. 14 illustrates the MC Method second step--determining
the pressure target based on using a colder than ambient soak
assumption.
[0022] FIG. 15 illustrates the MC Method third step, using the
pressure target from the second step in determining the expected
result and, if in excess of the target pressure, reducing the
target density and recalculating in an iterative manner to match
the pressure target at the final temperature.
[0023] FIG. 16 illustrates the results obtained from a 35 MPa Type
3 Tank Fill under a 35.degree. C. ambient with 5.degree. C.
Pre-cooled Hydrogen and 5 MPa Start Pressure. T.sub.final target is
69.2.degree. C., bounded by Hot Soak at 74.3.degree. C. and Cold
Soak at 62.3.degree. C.
[0024] FIG. 17 illustrates Result of 50 MPa Type 4 Tank Fill from
30.degree. C. Ambient with -15.degree. C. Pre-cooled Hydrogen and 2
MPa Start Pressure. T.sub.final target is 86.7.degree. C., bounded
by Hot Soak at 89.0.degree. C. and Cold Soak at 83.0.degree. C.
Note that the target pre-cooling temperature was -20.degree. C.,
which verified the difficulty in practice of keeping a specified
pre-cooling temperature.
[0025] FIG. 18 illustrates 70 MPa Type 4 Tank Test from 25.degree.
C. Ambient at 17 MPa Start Pressure with -7.5.degree. C.
Pre-cooling. T.sub.final target is 76.6.degree. C., bounded by Hot
Soak at 81.0.degree. C. and Cold Soak at 70.0.degree. C.
[0026] FIG. 19 illustrates Calculation of the Constants of the MC
Equation (Equation [11]) for the Type 3 Tank shows that the data
generated during verification filling under different conditions
compliments the data that was used to generate the constants
originally. The model is robust to different conditions.
[0027] FIG. 20 illustrates Error between T.sub.final as calculated
by the MC Method and the actual measured final temperature at the
end of the fill.
[0028] FIG. 21 illustrates a Sensitivity Analysis of Type 3 and
Type 4 Tanks to Input Errors Using the MC Method.
[0029] FIG. 22 illustrates a comparison of fueling methods showing
the impact of adding the MC Method to existing fueling
protocols.
[0030] FIG. 23 illustrates a relationship between nozzle
temperature deviation from pre-cooling temperature, flow rate,
pre-cooling temperature, and ambient temperature measured during
fill testing. This relationship is, of course, dependent on the
specific implementation of components for a given fueling station
or test stand.
[0031] FIG. 24 illustrates equations of a curve fit to the NIST
Hydrogen Property Database for hydrogen gas utilized for
determining internal energy (given temperature and pressure),
enthalpy (given temperature and pressure), temperature (given
internal energy and pressure), and density (given temperature and
pressure) of hydrogen gas.
[0032] It should be noted that these Figures are intended to
illustrate the general characteristics of methods, structure and/or
materials utilized in the example embodiments and to supplement the
written description provided below. These drawings are not,
however, to scale and may not precisely reflect the precise
structural or performance characteristics of any given embodiment,
and should not be interpreted as defining or limiting the range of
values or properties encompassed by example embodiments.
DETAILED DESCRIPTION
[0033] The goal of the methods and systems detailed in this
disclosure are to provide and utilize both a filling model and an
associated algorithm that can be used by a hydrogen tank filling
station or, indeed, any gas tank filling operation, to improve the
accuracy of the end-of-fill temperature and pressure for a range of
hydrogen tanks and a range of ambient and operating conditions.
Implementation of the methods and systems detailed below during
hydrogen tank refueling events will improve the efficiency,
accuracy and/or safety of the refueling operation by avoiding
overfilling and avoiding overheating the hydrogen tank.
[0034] Accurately estimating the end-of-fill temperature of a
refueling event is difficult, which is why communication refueling
has been developed, where temperature and pressure information is
directly transmitted to the hydrogen tank filling station via one
or more communication device(s) including, for example, the
Infrared Data Association (IRDA) interface detailed in SAE TIR
J2799, the disclosure of which is incorporated herein by reference,
in its entirety. The corresponding lack of such temperature and
pressure information is the reason that non-communication fueling
protocols requires large margin of safety, particularly given the
additional unknown parameters including, for example, the tank
type, the tank size, the aspect ratio, the number of tanks, hot or
cold soak status. Although full-communication fueling can be used
to provide the tank parametric data to the hydrogen tank filling
station, full-communications fueling adds more cost and complexity
to both the station and the vehicle and raises additional concerns,
particularly with regard to the use of in-tank sensors.
Accordingly, there remains a need for a method that provides for
sufficiently accurate predictions regarding the temperature of the
hydrogen in the tank during refueling without requiring
full-communication protocols and hardware.
[0035] In order to provide an accurate prediction of the
temperature of the gas, both the amount of energy that has been
transferred to the tank and the quantity of heat that has been
transferred from the gas to the wall of the tank are estimated.
Many studies have been conducted in trying to understand and
quantify the end-of-fill temperature. By modeling a hydrogen tank
as an open system in unsteady flow, as illustrated in FIG. 1, it is
possible to estimate the amount of energy that has been transferred
to a tank by measuring the enthalpy of the incoming hydrogen stream
and monitoring the temperature of the tank. For the purposes of
this disclosure, the control volume is defined as the boundary
between the gas and the tank liner with heat being transferred
through the boundary of the control volume and into the liner of
the tank. The total heat transfer in and out of the system, Q, is
reflected in Equations [1] and [2].
Q - W = .intg. m e h e m - .intg. m i h i m + .DELTA. E system ,
and setting m e = 0 and W = 0 [ 1 ] Q = - .intg. m i h i m + ( m 2
u 2 - m 1 u 1 ) cv [ 2 ] ##EQU00001##
[0036] where
[0037] Q=Total heat transfer in and out of the system (kJ) year
transfer out of the system is negative by convention)
[0038] W=work done to the system (kJ)
[0039] m.sub.e=mass exiting the system (kg)
[0040] h.sub.e=enthalpy of the fluid exiting the system (kJ/kg)
[0041] m.sub.i=mass entering the system (kg)
[0042] h.sub.i=enthalpy of the fluid entering the system
(kJ/kg)
[0043] m.sub.1=mass of the fluid in the control volume at the
initial state (kg)
[0044] m.sub.2=mass of the fluid in the control volume at the final
state (kg)
[0045] u.sub.1=internal energy of the fluid in the control volume
at the final state (kJ/kg)
[0046] u.sub.2=internal energy of the fluid in the control volume
at the initial state (kJ/kg)
[0047] cv designates the state of the control volume.
[0048] The enthalpy can be determined by measuring the temperature
and pressure of the hydrogen in the flow stream, preferably at a
point close to or at the tank inlet, with the mass flow into the
tank being measured or calculated from initial and final
conditions. To estimate the final temperature of the gas during or
after a refueling procedure, the actual heat transfer, Q, from the
gas into the tank wall needs to be estimated. Because Equation [2]
only gives information for the internal energy state of the tank a
tool, such as the National Institute of Standards and Technology
(NIST) Thermophysical Property Database, is used to look up the
temperature from the internal energy properties of the target gas,
e.g., hydrogen. A curve fit to the NIST data used here for internal
energy is illustrated in FIG. 2. The difference between the
adiabatic internal energy and the measured internal energy (u.sub.2
at measured temp and pressure) is the quantity of heat that is
transferred from the control volume, and can be determined from
test data.
u adiabatic = m 1 u 1 + .intg. m i h i m [ 3 ] m 2 = m 1 + .intg. m
i m = m 1 + m i [ 4 ] Q = m 2 ( u adiabatic - u 2 ) = m cv ( u
adiabatic - u 2 ) = m cv C v ( T adiabatic - T final ) [ 5 ]
##EQU00002##
[0049] where
[0050] u.sub.adiabatic=adiabatic internal energy--if there was no
heat transferred from the system (kJ/kg)
[0051] m.sub.2=m.sub.cv=end-of-fill mass of hydrogen in the control
volume (kg)
[0052] T.sub.adiabatic=adiabatic temperature--if there was no heat
transfer from the system (K)
[0053] T.sub.final=hydrogen temperature at the end of the fill
(K)
[0054] C.sub.v=Specific heat capacity of hydrogen at constant
volume (kJ/kgK)
[0055] FIG. 2 illustrates a Temperature v. Time curve for a
hydrogen tank refueling procedure in which the adiabatic
temperature, T.sub.adiabatic, is used in calculating the heat
transfer. The heat transferred from the hydrogen can be described
by Equation [5] as detailed above. This provides a tool for
analyzing actual test data to determine the amount of heat that has
been transferred from the hydrogen gas out of the control volume,
or into the structure of the tank. Note that the adiabatic internal
energy is based only on initial conditions of the tank (initial
mass of hydrogen and initial temperature) and the conditions of the
hydrogen delivered from the station (enthalpy and fill mass). There
is no dimension of time in the adiabatic condition, and so it is an
appropriate reference for results for all time periods. If a
reliable method to predict the heat transfer can be found, then it
should be possible to directly calculate the final state of the
hydrogen in the tank.
[0056] To calculate the end-of-fill gas temperature (T.sub.final),
it is necessary to estimate the amount of heat that is absorbed by
the tank wall. One method of calculating the total heat transfer is
to integrate the temperature distribution over the tank volume at
the end of a vehicle refueling procedure. In 2003, General Dynamics
conducted a series of tests aimed at understanding the relationship
between final tank temperature and filling time with the results
being reported in Eihusen, J. A., "Application of Plastic-Lined
Composite Pressure Vessels For Hydrogen Storage," World Hydrogen
Energy Conference, 2004 (Eihusen), the disclosure of which is
incorporated herein by reference, in its entirety. As detailed in
Eihusen, a series of filling tests were conducted while measuring
the temperature of the gas and of various locations in the hydrogen
tank wall.
[0057] These results indicated that during a refueling operation,
the heat transfer process was a process in which the temperature of
the outer surface of the tank did not rise, indicating that no
appreciable quantity of heat was being transferred through the tank
wall. Based on these results, General Dynamics proposed a heat
transfer model for predicting the temperature distribution within
the tank wall which was based on a Green's Function solution to the
general heat equation with a constant heat flux on the inside
surface and an adiabatic boundary on the outside surface. See
Equation [6]. A constant heat flux model showing temperature
distribution dependent on time with an adiabatic boundary condition
is illustrated in FIG. 4. Note that the assumption of no heat
transfer from the outside of the tank is conservative, meaning that
the actual final temperature in the tank will tend to be somewhat
lower than the final temperature calculated using this
assumption.
T ( x , t ) = q 0 L k { .alpha. t L 2 + 1 2 ( x L ) 2 - x L + 1 3 -
2 .pi. 2 m = 1 100 1 m 2 [ cos ( m .pi. x L ) ] - m 2 .pi. 2
.alpha. t L 2 } + T 0 [ 6 ] ##EQU00003##
[0058] where
[0059] T(x,t)=Temperature at liner depth=x, time=t
[0060] q.sub.0=Normalized heat flux of the liner (determined from
testing)
[0061] L=Thickness of the liner
[0062] k=Thermal conductivity of the liner
[0063] .alpha.=Thermal diffusivity of the liner
[0064] T.sub.0=Initial temperature of the liner
[0065] Inherent in this approach is the assumption that given an
initial set of conditions (hydrogen tank temperature, fuel gas
temperature and pressure) the fueling temperature result and the
temperature distribution result is purely dependent on time. The
amount of heat transferred to the liner during a refueling
procedure could be estimated by integrating the temperature
distribution over the volume of the liner. General Dynamics found
that for the given set of tests, this model predicted the final
tank temperature within 3K. The assumption of a constant heat flux
(or of temperature dependence only on time) is, however, both
interesting and problematic. Different initial conditions
(temperature of the tank, initial fill mass of the tank,
temperature and/or pressure of the incoming gas) will produce
different temperature gradients, and hence a different average heat
flux. Further, the heat flux would depend on the conditions of the
boundary layer between the gas and the wall--a change in the
velocity, density, or temperature of gas flow over the wall
resulting from forced and/or free convection inside the tank would
result in a corresponding change in the heat transfer. As these
conditions do tend to vary during the actual tank filling
procedure, the heat flux will also tend to vary over the course of
the filling procedure.
[0066] Furthermore, hydrogen refueling tests by the Japan
Automobile Research Institute (JARI) as reported in Hirotani, R.,
et al., JARI, "Thermal Behavior in Hydrogen Storage Tank for Fuel
Cell Vehicle on Fast Filling," World Hydrogen Energy Conference,
2006, the disclosure of which is incorporated herein by reference,
in its entirety, revealed a significant temperature distribution
within the gas itself, up to a 30K difference throughout the tank,
which would influence the heat flux. The presence of such
temperature distributions further complicates the analysis of data
as well because it renders it difficult, if not impossible, to know
precisely how accurately a particular temperature measurement taken
inside the tank represents the bulk properties of the tank.
[0067] Relative to the bulk temperature, if the thermocouple is
measuring a temperature that is warmer or cooler than the average
bulk temperature, the calculated values will obviously be less
accurate. Although it is customary to assume that temperature
measurements taken at or near the centerline of the tank will
represent the average (bulk) gas temperature, the magnitude of the
error(s) associated with this assumption are unknown in practice.
Data taken during the development of the method and system
disclosed herein reflected show unexplained errors of .+-.5K
between the thermocouple output and the expected bulk gas
temperature. Modeling efforts by the SAE TIR J2601 committee
similarly reflected errors of up to 9K between the modeled
temperature and the measured data despite the use of thermocouples
that have an accuracy of approximately .+-.1K. These real world
complications to the temperature gradient modeling make it
difficult in practice to apply a heat transfer model for directly
estimating the temperature distribution in order to calculate the
end-of-fill temperature (with reliance on one or more temperature
signal(s) during communication fueling operations introducing a
degree of uncertainty).
[0068] One objective of the JARI testing was to analyze the impact
of utilizing different refueling patterns while still keeping the
same overall fueling time. An interesting result of the JARI
testing indicated that, given the same initial conditions and a set
filling time, the temperature of the gas at the end-of-filling is
similar, regardless of the particular filling pattern used in
conducting the filling operation.
[0069] Analysis presented by St. Croix Research to the SAE TIR
J2601 Committee, Powars, C., "70-MPa Hydrogen Tank Filling Model
and Test Data Observations--SAE TIR J2601 Modeling Sub-Team
Meeting, Sep. 15, 2008," St Croix Research, 2008, the disclosure of
which is incorporated herein by reference, in its entirety, also
shows that the outer tank temperature does not rise significantly
during a vehicle refueling.
[0070] While complex heat transfer models and analysis were
previously proposed to explain the results of these tests, the
innovative approach described herein is based on analyzing tank
refueling processes using a simple lumped heat capacity model, the
development of which is described in more detail below and further
illustrated in the associated figures. The utility and
applicability of this novel solution can be confirmed by using the
JARI, General Dynamics, SAE and St. Croix Research results to show
that the simplified model based on lumped heat capacitance detailed
herein models and predicts the heat transfer characteristics of the
system with sufficient accuracy. Additional testing and analysis of
fueling data was conducted and the results further verified that
this new method of using this lumped heat capacitance model is
sufficient for accurately describing the thermodynamics of hydrogen
tank refueling systems and improving the associated refueling
processes.
[0071] Consider a tank that has just completed a vehicle refueling
in a short period of time. As illustrated in FIG. 5, the inside of
the tank is much hotter than the outside of the tank, due to the
conversion of pressure energy to sensible energy of the high
pressure hydrogen that was just recently injected into the tank.
Calculating the actual temperature distribution is difficult
because 1) there is a temperature distribution between the hydrogen
in the tank and the liner due to boundary conditions, 2) there is a
temperature distribution through the liner, contact resistances
between layers, 3) there is a temperature distribution through the
various layers of the tank and 4) there is a temperature
distribution between the outside of the tank and the environment
due to the outer boundary conditions. And as discussed previously,
there may also be a temperature distribution within the hydrogen in
the tank itself on the order of 30K. Each layer has a different
specific heat capacity that might also be dependent on temperature,
and each layer is composed of a different mass. Given all of these
complexities, it is exceedingly difficult, if not impossible, to
calculate the precise temperature distributions through the wall of
the tank.
[0072] The biggest difficulty in considering a heat transfer model
based on a precise calculation of the temperature distribution in
the wall of a hydrogen tank is that it requires a solution to the
temperature distribution for the entire time domain of a refueling
event, a solution which is difficult to achieve in practice.
Accordingly, the method utilizes a combined mass and specific heat
capacitance model, as shown in FIG. 5, which illustrates a
simplified section of a tank wall having an imaginary
characteristic volume defined by its mass (M) and specific heat
capacity (C) and an adiabatic external boundary. Tke temperature of
the characteristic volume will be the same as the temperature of
the gas, for example, hydrogen.
[0073] This section of the tank wall, the characteristic volume,
will have a combined mass and specific heat capacity, MC (kJ/K).
Note that the characteristic volume and the associated MC are
mathematical constructions only. Given knowledge of a tank design
and the materials used in the tank construction, it is possible to
provide a reasonable estimation of the MC value. In the method
disclosed herein, however, there is no need to calculate the exact
mass and specific heat capacity of the tank because the
characteristic volume simply acts as a heat sink, allowing its
characteristics to be used in predicting the thermal behavior of
the tank system.
[0074] In applying the method the temperature of the characteristic
volume is set to be equal to the temperature of the hydrogen in the
tank at the end of the vehicle fill. This means that the
characteristic volume has both high thermal conductivity and high
convective heat transfer coefficient. In addition,
Q.sub.Environment=0, meaning that no heat is transferred out of the
characteristic volume during the fueling (adiabatic boundary). As
discussed supra, there is very little heat transfer to the
environment during the refueling operation, allowing this component
of the heat transfer to be ignored. In an illustrative example, the
heat transfer equation is solved for the target or preferred
end-of-fill condition of, for example, a fill time of 2 or 3
minutes, plus some adjustment for longer fill times as deemed
necessary. For the purposes of this disclosure, a target fill time
of 3 minutes has been used, but those skilled in the art will
appreciate that the method can be easily utilized for longer or
shorter fill times.
[0075] When applying a characteristic volume in this manner, the
heat that is transferred from the hydrogen mass, m.sub.cv, into the
characteristic volume can be described by the temperature rise
during the fueling of the characteristic volume with a combined
mass and specific heat capacity of MC.
Q=MC(T.sub.final-T.sub.initial) [7]
[0076] where
[0077] MC=Combined Mass and Specific Heat Capacity of the
Characteristic Volume (kJ/K)
[0078] T.sub.final=Temperature at the finish of the refueling
(K)=
[0079] T.sub.initial=Temperature at the beginning of the refueling
(K)
[0080] By applying an energy balance across the boundary of the
control volume, and combining equation [5] (energy transferred from
the hydrogen) with equation [7] (energy transferred to the
characteristic volume) results in:
Q = m cv ( u adiabatic - u final ) energy transferred from the
hydrogen control volume [ 5 ] Q = M C ( T final - T init ) energy
transferred to the characteristic volume [ 7 ] M C = m cv ( u
adiabatic - u final ) ( T final - T init ) [ 8 ] ##EQU00004##
MC can then be determined directly from test data for a particular
refueling procedure by using Equation [8], which is the ratio of
the heat transferred from the hydrogen to the temperature change of
the characteristic volume. For Equations [7] and [8],
T.sub.init=T.sub.initial. The temperature, pressure and MC behavior
associated with a 3-minute fill of a Type 3 tank is illustrated in
FIG. 6. As reflected in the graph, the MC is 62 at the end-of-fill
point, but then increases over time as the tank cools. This trend
of MC over time can, in turn, be used in predicting the result of
longer filling procedures. Once an MC is known for a given set of
initial conditions, it can be used for directly calculating the
final temperature of refueling event.
Q = m cv ( u adiaatic - u final ) [ 5 ] Q = m cv C v ( T adiabatic
- T final ) , another identity of equation 5 [ 9 ] Q = M C ( F
final - T init ) Combining equations [ 9 ] and [ 7 ] we get : [ 7 ]
T final = m cv C v T adiabatic + M C T init M C + m cv C v ,
combining equations [ 7 ] and [ 9 ] [ 10 ] ##EQU00005## [0081]
where [0082] C.sub.v=specific heat capacity of hydrogen at constant
volume, kJ/(kgK)
[0083] Equation [10] can then be used to calculate the expected
final temperature of a hydrogen tank refueling just as a fill has
started. The MC parameter and m.sub.cv (the end-of-fill mass in the
control volume) are transmitted to the station. In a nonlimiting
example, the MC parameter and m.sub.cv are transmitted by RFID,
through the SAE TIR J2799 IRDA interface, or via an identification
number that corresponds to entries in a database that is readily
accessible to the hydrogen tank filling station. The hydrogen tank
filling station can calculate T.sub.adiabatic from m.sub.cv and
parameters including 1) the initial pressure of the tank receiving
the hydrogen (e.g., the vehicle's tank), 2) the initial temperature
of the tank receiving the hydrogen (assuming ambient conditions
plus some differences due to the possibility of a hot or cold tank
as discussed in SAE TIR J2601) and 3) the enthalpy of the delivered
hydrogen, which is a function of the expected average temperature
and pressure of the delivered hydrogen (further description is
given in the Appendix provided in FIG. 24).
[0084] Certain characteristics of the MC Method make it
particularly useful for gas delivery systems. For example, a
particular tank configuration will have a characteristic curve of
MC v. fill time from which adjustments can be made to compensate
for a range of initial conditions. Utilizing the MC model avoids
the need to address all of the intricacies of the temperature
distribution of the wall of the tank, especially over a time scale
associated with typical hydrogen tank refueling procedures, e.g.,
two to three minutes or more.
[0085] MC is not a direct physical constant such as the mass and
the specific heat capacity of the tank and liner material but is
rather it is a composite value, similar to an overall heat transfer
coefficient, that encompasses heat transferred to tank valve
assemblies and piping as well as heat transferred to the hydrogen
comprising the initial gas volume inside the tank being filled.
Systems with slower heat transfer characteristics (convection or
conduction) will tend to result in lower values of MC (such as Type
4 tanks) while systems with faster heat transfer characteristics
(convection or conduction) will tend to result in higher values of
MC (such as Type 3 tanks). Although MC is a function of a number of
different parameters including, for example, time, fill conditions,
tank materials, tank configuration, etc., for a given tank, fill
time and set of fill conditions, MC will be constant. The trend in
the MC value over both time and under different fill conditions can
be predicted and, in turn, utilized for adjusting the hydrogen tank
filling procedures to improve efficiency while maintaining desired
safety margins.
[0086] Based on observations of several sets of test data coupled
with the use of multiple linear regression for evaluating the
significance of various parameters, many possible physical models
were considered for describing the MC v. time curve and also
describing the changes in initial conditions that were tested. In a
non-limiting example, one model is represented by Equation [11], as
shown below:
M C ( U , t ) = C + Aln ( U adiabatic U initial ) + g ( 1 - - k
.DELTA. t ) j [ 11 ] ##EQU00006##
or, in an alternative non-limiting example, Equation [11]', as
shown below:
M C ( U , t ) = C + A ( U adiabatic U initial ) + g ( 1 - - k
.DELTA. t ) j [ 11 ] ' ##EQU00007## [0087] where C, A, g, k and j
are constants derived from characterization testing [0088]
U.sub.adiabatic is the adiabatic internal
energy=m.sub.cvu.sub.adiabtic [0089] U.sub.initial is the initial
energy=m.sub.initialu.sub.initial [0090] .DELTA.t is the difference
in time between the normally defined end-of-fill time (e.g., 3
minutes) and the end-of-fill time that achieves the desired final
temperature.
[0091] In the context of Equation [11], C is a constant that
represents a minimum heat capacity of, for example, a 2- or
3-minute fill, A is a constant representing an adjustment to the MC
corresponding to the initial fill conditions and pre-cooling amount
and constants g, k, and j are, if necessary, utilized within the MC
Method to provide the ability to adjust the resulting MC by
extending the fill time beyond 2 or 3 minutes, so that T.sub.final
can be optimized around a desired temperature. However, it is to be
understood that those skilled in the art will appreciate that there
are many possible models that can be developed to predict the trend
of MC with time, internal energy, pre-cooling temperature, etc. The
MC Method is not intended to, and does not attempt to, perfectly
describe the physics but instead provides an analytical engineering
tool that can be used for predicting the temperature outcome of a
particular filling procedure by approximating the equivalent heat
mass of the system.
[0092] One way to check a new model is to verify that the model is
capable of describing or predicting phenomena documented in
previous literature. The Society of Automotive Engineers (SAE)
conducted several sets of hydrogen tank fill testing at Powertech
during the development of SAE TIR J2601, in support of, and to test
the modeling efforts that were being conducted to build the
refueling tables in SAE TIR J2601. By using the tables provided in
SAE TIR J2601 as a set of test fills and plotting the MC of each
fill versus the fill time using Equation [8], the results fall in a
distinct pattern as illustrated in FIG. 7.
[0093] This result is encouraging for several reasons, including
but not limited to 1) because the MC Method describes the actual
results of the SAE TIR J2601 tests quite well, suggesting that the
model accurately represents the physics of hydrogen tank filling,
and 2) because the entire set of tables in SAE TIR J2601 can be
approximated using this single equation. This result indicates that
the equation utilized in the MC Method can be used to describe the
MC v. Time over a wide range of conditions, and can be used in
place of the several sets of tables defined in SAE TIR J2601, which
require interpolation between the listed data to find the
appropriate pressure ramp rate and end of fill pressure target.
Because the MC Method can adjust the fill time to match a desired
final tank temperature, it can be used for any station
configuration. This releases the fill protocol from reliance on the
rigid "Type A, B, C, D" station type designations of SAE TIR J2601,
because the resulting fill temperatures can be derived for a wide
range of station conditions. Indeed, by using the coefficients of
the MC v. time curve utilized in the MC Method, a hydrogen tank
filling station can directly calculate the expected end-of-fill
temperature (T.sub.final) using Equation [10].
[0094] As will be appreciated by those skilled in the art, using
the SAE TIR J2601 tables to calculate a characteristic MC curve
provides a characteristic curve that corresponds to a "default SAE
TIR J2601 tank." Accordingly, when this MC curve is used in
Equation [10], the fill results are substantially the same as those
represented in the SAE TIR J2601 tables. The MC Method, however, is
intended to provide for fill times that are both shorter than
predicted from the SAE TIR J2601 tables and provide for improved
fill quality. In order to achieve this result, the MC Method
incorporates the specific characterized MC curve corresponding to a
specific tank.
[0095] A set of fill tests were conducted at Powertech during Feb.
1-6, 2010 and Aug. 23-30, 2010 in order to characterize a 171L Type
3-35 MPa tank, an approximately 109L Type 4-70 MPa tank, and the
same Type 4-70 MPa tank filled to 50 MPa, and a 34L Type 3-70 MPa
tank. Each tank was tested using pre-cooled and non-pre-cooled gas,
at 25.degree. C. ambient, and 2 MPa starting pressure or 1/2 tank
starting pressure. Each tank was filled in approximately 1 to 3
minutes at the given conditions and data recorded for 1 hour
following. Each tank was then defueled and allowed to soak at the
ambient temperature until the next test the following day.
[0096] Using Equation [8], the MC v. time was plotted for each
fill, as shown in FIG. 6. All of the tank fills follow a similar
pattern of MC v. Fill Time as shown in FIG. 7. The resulting curve
corresponds to the tank characteristic(s) for a given tank under a
given set of conditions. To find the coefficients used in Equation
[11], the MC for each end-of-fill at 3 minutes was plotted against
the adiabatic internal energy divided by the initial internal
energy, as shown in FIG. 9. The slope and intercept of the linear
best fit line give the coefficient A and the constant C
respectively. The .DELTA.MC v. .DELTA.time, that is
(MC.sub.(t-180s)-MC.sub.(180s)) v. (t-180 s), is then plotted as
shown in FIG. 10, and a best fit model is used to determine the
coefficients g, k and j. These coefficients can then be used to
describe how much heat is absorbed by the tank in the time beyond
the typical fill time and are particularly useful under conditions
in which the ambient temperature is too warm and/or the pre-cooling
temperature is too warm to achieve an end-of-fill temperature of
less than 85.degree. C. with a refueling time of 3 minutes or
less.
[0097] To use the MC parameters for improving the performance of a
hydrogen tank filling station, a fueling protocol needed be
developed. A fueling protocol should provide safe, high state of
charge (SOC) fills, for a broad range of ambient conditions and
initial fill conditions. Comparing the current fueling standards
with the actual operating ranges of existing hydrogen stations, as
illustrated in FIG. 11, it is clear that the current refueling
standards do not satisfy a broad range of station fuel delivery
operating conditions. Further, should a vehicle manufacturer or
modifier introduce a tank designed to operate at another pressure
of, for example, 50 MPa, the fueling standard(s) would have to be
rewritten to accommodate this modification.
[0098] In order to fully utilize the MC Method at an actual fueling
station, the MC parameters must be communicated to or determined by
the station in some manner. This data collection could be achieved
in a number of ways. In a non-limiting example, RFID, or even the
IRDA interface defined in SAE J2799, may be used to transmit the MC
parameters from the vehicle to the station. There is a working
group within the California Fuel Cell Partnership that is
developing a Hydrogen Vehicle Authorization System (HVAS) to be
used for confirming that a vehicle is authorized to fuel (OEM
vehicle or a conversion that meets safety requirements). The HVAS
specifications and device are still under development but it is a
candidate for communicating the MC parameters to the station,
either directly through the device or, alternatively, by matching
the identified vehicle to a database from which these parameters
could be retrieved.
[0099] FIG. 12 shows both the vehicle side and station side
information that may be used to fuel a vehicle based on the MC
Method. The station will have access to both the vehicle side
information through, for example, HVAS and station side information
through direct measurement. The process the station goes through to
determine the appropriate fueling speed and end-of-fill pressure
target is very similar to that used in building the lookup tables
in SAE TIR J2601. The difference is that the assumptions utilized
in J2601 are worst case, and based on boundary condition tanks,
whereas in conjunction with the MC Method, the station is given
some basic parametric data and tailored assumptions for filling the
particular vehicle.
[0100] The MC Method can also readily accommodate the communication
of specific modification to the fill protocol preferred or
specified by the OEM. For example, if an OEM imposes or suggests a
maximum fill rate, develops a tank system in which the maximum
temperature can exceed 85.degree. C. or allows fueling to 103% SOC
(if inside of the Maximum Allowable Working Pressure (MAWP)),
parameters related to the OEM's design or operating limits and/or
preferences can be provided to the hydrogen tank fueling station
for modifying the fill protocol accordingly. This flexibility puts
additional control of the outcome of the vehicle refueling squarely
into the hands of the OEM, so that fill stations utilizing the MC
Method can adapt the fill protocol to accommodate the particular
vehicle and thereby permit a broader range of OEM designs.
[0101] In an embodiment, when applying the MC Method, the fueling
process can comprise two discrete steps. In the first step,
parametric data is used to determine an appropriate fueling fill
rate, i.e., one that will not overheat the gas in the tank. During
the second step, the fueling fill rate is used to determine a
target end-of-fill pressure that will keep the system pressure
within the target pressure ranges. These two steps are explained
below in more detail.
[0102] In order to determine the appropriate fueling rate for the
projected fill operation, the hydrogen tank filling station takes
into consideration both the capabilities of the vehicle tank system
and its own capabilities to deliver the fuel under the current
conditions.
[0103] The limits of refueling, as defined in SAE TIR J2601 and TIR
J2579 are 85.degree. C. and 125% of the NWP for average gas
temperature and pressure, respectively. In an illustrative example,
the station makes an assumption about the average gas temperature
inside the tank, based on measuring the ambient air temperature and
optionally adding a margin for a hot soak condition (e.g., the
vehicle has been parked in an environment that is hotter than
ambient, such as a hot garage or parking lot). The station also
determines the approximate initial SOC of the vehicle, using the
temperature assumption and by dispensing a small amount of fuel to
the tank to equilibrate the hose pressure to the tank pressure.
Based on the tank pressure and vehicle side information, the
station can estimate how much hydrogen (mass) must be delivered to
the vehicle to achieve the desired SOC and, utilizing an estimate
of its pre-cooling capability, the station can calculate the
average enthalpy that will be delivered to the vehicle's tank
system during the fill operation. Working from this information,
the station can then determine how quickly to fill the vehicle
while maintaining the requisite safety margin.
[0104] As explained supra, the primary MC parameter is based on a
target fueling time with additional parameters being used to
account for the initial SOC and/or fueling times that exceed the
target fueling time. Starting with these targets, the station
analyses an initial fill protocol to determine if the fill can be
successfully completed, i.e., end-of-fill temperature within
specification. If it is determined that the initial fill protocol
cannot be successfully completed, an iterative process is initiated
to determine an appropriate fueling time. For example, if the
fueling operation can be conducted in the target time without
exceeding any temperature limits, the station will initiate
fueling. If, however, the initial fill protocol would cause a
temperature limit to be exceeded, the projected fueling time will
be increased by some increment (e.g., 0.1, 1, 5, 10 seconds, etc.)
and the new MC value will be calculated. This incremental increase
of the fueling time will continue until a fueling time is
identified that will result in end-of-fill conditions that are
within specification, e.g., the end-of-fill gas temperature is less
than 85.degree. C. This process is shown in FIG. 13. The output of
this Step 1 is the T.sub.final(Hot Soak Bound) and the fueling or
fill time. In an embodiment, the appropriate fueling time can be
continuously calculated throughout the fill procedure based on the
actual enthalpy delivered to the vehicle. Accordingly, even though
the fueling time calculated at the beginning of the fill should be
a good approximation, the fueling time (or rate of pressure rise
during the fill) can be adjusted as necessary utilizing a feedback
loop based on the actual fill conditions as they occur.
[0105] For the dispenser to make the assumption that the upper
bound of gas temperature inside the tank is ambient T plus a
.DELTA.T hot soak, it must know that the vehicle has not been
refueled in the recent past. If it cannot know this information,
then it should make a more conservative assumption, and determine
the fueling speed based on an empty or nearly empty tank. Then,
even if the vehicle was recently refueled, the fueling speed will
not overheat the tank.
[0106] If the recent fueling history of the vehicle can be
determined, a less conservative fueling speed can be utilized,
potentially shortening the fueling time considerably. There are a
number of approaches that can be utilized for determining the
recent fueling history of the vehicle. A non-limiting example is
for the HVAS RFID tag to be time stamped each time the vehicle is
fueled. The dispenser can then read this time stamp each time the
vehicle is fueled and determine whether to use a conservative
fueling speed if the time stamp indicates a recent refueling, or a
less conservative fueling speed based on the actual starting
pressure in the tank if the time stamp indicates refueling has not
occurred recently.
[0107] Once the appropriate fueling time has been determined, the
next step of the MC Method is to determine when, or at what
pressure, to stop the fill operation. The process used by the
station in this second step is similar to that used in the first
step except that the station assumes the gas temperature inside the
tank at the beginning of the fill is below the ambient temperature,
i.e., a cold soak condition, which includes the possibility that
the tank has been soaked in an air conditioned garage, or that the
ambient temperature is rising and the internal gas temperature lags
the ambient. There is also the factor of driving that may be
considered in which the gas temperature inside the tank has been
reduced as a result of the decrease in pressure as the hydrogen was
consumed. The MC Method can be used to estimate the average
temperature of the MC and hydrogen gas during defueling using
Equation [12]
u adiabatic ( t ) = U ( T ColdSoak , P NWP ) - m add ( t ) h exit m
initCold [ 12 ] ##EQU00008## [0108] where [0109] M.sub.add=mass
exiting the hydrogen tank in time t [0110]
m.sub.add=m.sub.cv-m.sub.initCold=mass to be added during the
vehicle refueling calculated in the [0111] MC Method to achieve
100% SOC [0112] h.sub.exit=average enthalpy of the hydrogen exiting
the tank [0113] m.sub.init Cold=mass in the tank just before
refueling [0114] t=time it would take to empty the tank from
P.sub.NWP to the starting fill pressure P.sub.init
[0114] t = m add m . [ 13 ] ##EQU00009## [0115] where [0116] {dot
over (m)}=flow rate of hydrogen during defueling (g/s) [0117]
T.sub.ColdSoak=assumed temperature of the vehicle tank before
defueling
[0117] T.sub.ColdSoak=T.sub.ambient-.DELTA.T.sub.Cold [14]
And combined with Equation [11] where the T.sub.adiabatic is
determined by a curve fit to NIST data as before, then T.sub.final
is the average temperature of the MC and the gas in the tank.
T FinalDefuelCold = m cv C v T AdiabaticCold + MC ( t ) T ColdSoak
MC + m cv C v [ 15 ] ##EQU00010##
[0118] The appropriate .DELTA.T.sub.Cold parameter, and the
defueling mass flow rate {dot over (m)}, will typically be
determined by the OEM and will be provided as part of the vehicle
side information transferred through HVAS or otherwise made
available to the filling station.
[0119] Once the initial conditions have been determined, the
station can calculate how much mass must be added to the tank to
reach the target density of 100% SOC. If the station has an
accurate flow meter, it can simply integrate the mass flow during
the fill and stop when the target mass has been achieved, however,
the application of a flowmeter in this capacity might have its own
challenges. A second option is to calculate a pressure target
utilizing the same set of equations as in Step 1. T.sub.final can
be calculated based on the fueling time of Step 1, and then the
P.sub.target value can be calculated based on the pressure that, in
conjunction with T.sub.final, provides a 100% SOC target
density.
[0120] This process can be more easily understood by utilizing the
equations shown in FIG. 14. It is important to note that the
pressure target can be continuously calculated throughout the fill
procedure based on the actual enthalpy delivered to the vehicle.
Accordingly, even though the pressure target calculated at the
beginning of the fill should be a very good approximation, the
pressure target utilized in stopping the fill can be adjusted as
necessary based on the actual fill conditions as they occur. The
output of this Step 2 is the P.sub.Target.
[0121] In the case of a fill with communications, the initial
temperature can be measured directly by the station. Because this
initial temperature is a settled temperature, i.e., a temperature
not subject to the dynamic changes associated with vehicle fueling,
it is typically reliable. In such cases, the T.sub.init is simply
the measured initial temperature and the hot soak and cold soak
assumptions detailed above need not be considered.
[0122] During the fill testing conducted during development of the
MC Method, a "Target T.sub.final" value was calculated in order to
evaluate any errors between the expected result and the actual
result. This "Target T.sub.final" is shown in FIGS. 16-18 and FIG.
20 to demonstrate the accuracy of the MC Method. In a normal
"ID-Fill," Step 3 is unnecessary--the station does not need to
calculate an expected result as the fill protocol is fully defined
by Step 1 and Step 2.
[0123] Using the fill rate from Step 1, and the Pressure Target
from Step 2, the expected T.sub.final can be calculated. Because
the Pressure Target calculated in Step 2 is usually lower than the
Pressure Target that was assumed in Step 1, the resulting fill will
tend to exhibit a slightly lower SOC % which, in turn, indicates
that the gas density target needs to be reduced to match the
Pressure Target at a higher T.sub.final than was calculated in Step
2. Because a change in additional mass of hydrogen added affects
the T.sub.adiabatic, for greater precision it is necessary to
complete the outlined calculations in order to determine the
expected T.sub.final and SOC % target.
[0124] As will be appreciated by those skilled in the art, the
utility and flexibility of the MC Method provides many
opportunities for customization and refinement to encompass, for
example, fueling times of less than 3 minutes for tanks that start
filling at high SOC.
[0125] To confirm the MC parameters calculated according to the
procedures defined supra, and to confirm the accuracy of using
these parameters in the filling algorithm detailed supra, a fifth
fueling test was conducted for each of the previously tested tanks
using conditions of ambient temperature, initial fill amount, and
pre-cooling temperature that were different than the conditions
used in characterizing the tank. Using the algorithms discussed
supra and illustrated in FIG. 13, the expected final temperature
T.sub.final was calculated for fills conducted at 35 MPa, 50 MPa
and 70 MPa. For the Hot Soak margin of safety to overheat,
T.sub.init=Ambient Temp+7.5.degree. C. was used, for the Cold Soak
margin of safety to overfill, T.sub.init=Ambient Temp -10.degree.
C. was used. For the target T.sub.final, T.sub.init=Ambient Temp
was used in the algorithm illustrated in FIG. 15.
[0126] The results of a 35 MPa Type 3 Tank Confirmation Test are
illustrated in FIG. 16. Although the original targets were set for
delivering 0.degree. C. gas, the hydrogen filling station being
used for the evaluation was actually delivering nearly 5.degree. C.
gas, which would be outside of the SAE TIR J2601 tolerance of
0.degree. C..+-.2.5.degree. C. for a Type C station. This
demonstrates one of the practical challenges of defining a tight
tolerance on the pre-cooling temperature--it is actually difficult
to achieve and/or maintain, even in test conditions. In light of
the noted capabilities of the hydrogen filling station, the targets
were adjusted for using 4.8.degree. C. as the temperature of the
delivered gas, the 35 MPa tank fill actual temperature measurement
was within 1K of the calculated T.sub.final. Further, although fill
completion was targeted for 180 seconds, the actual fill was
finished at 196 seconds. As a practical measure, in order to
achieve an optimum fill time the Hot Soak Bound should be set at
85.degree. C., however, because the test was predicated on a
3-minute fill target, the Hot Soak Bound is less than 85.degree. C.
The MC Method algorithm can be further refined to improve
performance for fill times of less than 3 minutes.
[0127] The results of a 70 MPa Type 4 Tank Filled to 50 MPa
Confirmation Test are illustrated in FIG. 17. In this instance,
although the pre-cooler was set for -20.degree. C., it was
determined that the pre-cooler was actually delivering
-14.8.degree. C. gas on average. This result once again reflects
the actual difficulty of meeting SAE TIR J2601 tolerances of
-20.degree. C.+/-2.5.degree. C. for a Type B station. In light of
the observed performance, the temperature targets were adjusted to
reflect what -15.degree. C. pre-cooling targets would have been,
given the same conditions otherwise. Although this rendered the Hot
Soak bound high at 89.degree. C., this deviation is a relic of the
pre-cooling temperature being out of specification.
[0128] Also noted were changes in temperature in the tank measured
after the end-of-fill. These post-fill deviations represent a
practical source of error in temperature measurements in a hydrogen
tank that may result from, for example, thermocouple placement,
temperature gradients within the tank and/or time lag. Given these
errors, however, the actual fill result was still remarkably close
to the target, further validating the model. Additionally,
85.degree. C. was not utilized as a stop point in these fills,
thereby allowing the tanks to reach temperatures slightly above
85.degree. C. These minor temperature deviations were not
considered problematic because transient temperatures above
85.degree. C. are generally known and allowed pursuant to SAE
J2579, the disclosure of which is incorporated by reference, in its
entirety.
[0129] The results of a 70 MPa Type 4 Tank Confirmation Test are
illustrated in FIG. 18. As reflected in the illustrated data, the
70 MPa tank test temperature result was an essentially perfect
match for the calculated T.sub.final Target.
[0130] Comparing the data obtained from the 4 test fills used to
generate the constants of Equation [11] to the data generated in
the fifth verification fill and additional verification fills, the
results reinforce the concept that the MC is characteristic for the
tank and can be used to predict the fueling result. This is
demonstrated in the graphs illustrated in FIG. 19 in which the data
generated during the Type 3 Tank confirmation tests detailed above
is consistent with the data used in determining appropriate values
for the various constants utilized in Equation [11]. These results
demonstrate that the MC Method is sufficiently robust to be applied
confidently across a range of tank configurations and operating
conditions.
[0131] Looking at the error from all of the fills conducted, as
illustrated in FIG. 20, it is apparent that the MC Method yields
very accurate results for Type 3 and Type 4 tanks, typically
falling within a range consistent with that expected from
variations in thermocouple placement and/or time lag errors. As
shown in FIG. 20, the MC Method Model error is the difference
between T.sub.final as calculated by the MC Method, and the actual
final temperature result measured at the end of the fill procedure.
The actual pre-cooling temperature of the station was used as the
input to the enthalpy calculation rather than the pre-cooler set
point, for the reasons described supra. Known or suspected sources
of error include, for example: [0132] errors of the calculation in
average enthalpy used, [0133] calculation in mass of hydrogen
delivered, [0134] measurement of ambient temperature, [0135]
measurement of initial tank pressure, [0136] measurement of final
tank pressure, [0137] measurement of final tank temperature
(thermocouple placement, lag, standard error) [0138] calculation of
the MC from the best-fit coefficients, and [0139] difference
between actual fill time and expected fill time (due to station
bank switching, flow differences, etc.), [0140] heat transfer in or
out of the hydrogen stream after the station enthalpy measurement,
and/or [0141] differences in the actual tank temperature from the
assumed ambient temperature (hot spots, cold spots, etc.)
[0142] Given all of these possible sources of error, it is
remarkable that the data generated during testing suggests that a
lumped heat capacity model can achieve a standard deviation of
errors of 0.6K for Type 3 Tanks and 2.4K for Type 4 Tanks. The
"Definition Error" as shown in FIG. 20 removes the error in
calculating enthalpy, calculating mass, and calculating MC
coefficients by using the test data to determine the actual heat
transfer, the actual average enthalpy of the fill, and the actual
MC value, and using those to calculate T.sub.final. This removes
substantially all of the errors and approximations attributable to
the calculations of the MC Method itself, leaving only the
measurement errors as the source of error. This has a standard
deviation of 0.3K for the Type 3 tank and 1.3K for the Type 4 tank.
The remaining portion of the errors is likely a result of
measurement errors, thermocouple lag and/or differences between the
assumed and actual conditions (such as cold spots in the tank after
defueling). It was noted that as the pace of the testing increased
the magnitude of the errors also tended to increase, possibly as
the result of differences between the assumed and actual conditions
including, for example, residual cold spots remaining from the
defueling operations conducted between filling tests.
[0143] A sensitivity analysis of the MC Method to variations in
input errors was conducted to examine the correspondence between
known levels of input errors and the resulting output errors. As
reflected in the data presented in FIG. 21, the MC Method was
relatively resistant to input errors with Type 3 tanks being more
sensitive to variations in the initial temperature measurements
while Type 4 tanks are more sensitive to variations in the
temperature measurement of the flow stream at the station. 10K
errors in the initial temperature measurement leads to 6K errors in
T.sub.final for both Type 3 and Type 4 tanks. 10K errors in the
hydrogen temperature measurement at the station (used for the
average enthalpy approximation) lead to T.sub.final error of 6K for
Type 3 tanks and 8K for Type 4 tanks. 10% errors in the calculated
MC coefficients lead to errors of around 3K (and 3K represents
approximately a 1% error in the density of hydrogen). These results
demonstrate that the MC Method has significant robustness to
accurately describe vehicle fueling over a range of conditions and
suppress the effect of input errors.
[0144] As detailed above, utilizing the MC Method for refining
Fueling Protocols can improve fueling performance. Although an ID
Fill fueling protocol was discussed supra, the MC Method may also
be applied to conventional non-communication fueling operations, as
well as full communication fueling operations, as currently defined
in SAE TIR J2601. A comparison of fueling methods is shown in FIG.
22, which highlights the benefits that could be expected to flow
from incorporating the MC Method into all three types of fueling
(i.e., ID Fill, Non-Communication and Full-Communication). These
benefits are further elaborated upon in the discussion provided
infra.
[0145] In an ID Fill configuration, the fueling process is better
adapted to the tank that is being fueled, thus tending to provide
reduced fueling time and increased SOC within the bounds of the
uncertainties of the initial conditions of the tank and the
measurements at the station. The fueling process is also better
adapted to the station's real time capabilities, thereby increasing
operational flexibility and avoiding the rigid, preset, tightly
bounded temperature requirements corresponding to the various
station types as defined in SAE TIR J2601. The MC Method allows the
filling process to self-adjust to the current fueling capabilities
of the station, thereby providing the potential for simpler, more
flexible and less costly hydrogen filling stations. The flexibility
of the MC Method allows a hydrogen filling station to be "tuned" to
the current operating environment which, in turn, may allow for
increased pre-cooling temperatures while still maintaining
generally acceptable fueling times under most conditions. The
ability to run at higher pre-cooling temperatures can improve
station efficiency, lower costs and maintain customer
satisfaction.
[0146] Fueling processes incorporating the MC Method as detailed
supra could eliminate the need for the look-up tables currently
utilized for non-communication fueling in accord with SAE TIR
J2601, resulting in the same benefits as outlined above. The
non-communication fueling operations could include calculations of
the MC Parameters of the boundary condition tanks utilized in
building the non-communication look-up tables. When operating at
the Type A (-40.degree. C.) or Type B (-20.degree. C.) pre-cooling
temperatures, the resulting range of fueling rates and pressure
targets would be expected to be substantially the same, if not
identical, to those defined in the look-up tables.
[0147] The flexibility of the MC Method in addressing variations in
temperature and pressure would reduce or eliminate the need for
rigid definitions of Station Types as currently applied and would
allow each station to operate more efficiently for its current
environment, and to dispense fuel at improved rates, regardless of
its pre-cooling temperature. Conversely, non-communication
processes as defined in SAE TIR J2601 must operate within very
tight pre-cooling tolerances, and if it falls outside them, cannot
dispense fuel (resulting in unhappy customers) until its margins
are back within the specified range(s).
[0148] The MC Method fueling process can also be utilized with full
communication fueling, resulting in a number of benefits. SAE TIR
J2601 currently defines two types of communication fueling
including 1) a Default method in which fueling rates are the same
as the non-communication fueling rates defined in the look-up
tables and 2) an Alt Method in which a more aggressive fueling rate
can be utilized in those instances in which a vehicle Temperature
Signal can be utilized in a feedback loop to regulate the fueling
rate in order to suppress or avoid an overheat condition. With the
MC Method, the fueling rate is determined at the beginning of the
fill, just as described above, and is also checked during the fill
based on the actual enthalpy of hydrogen delivered during the fill.
With communications fueling, the initial and transient conditions
can be more tightly defined, giving even better results.
Incorporation of the MC Method would mean that the Default and Alt
Methods would no longer be needed--a single communications fueling
protocol could be defined and it would be adapted for the vehicle
being fueled and the fueling conditions.
[0149] From a safety standpoint, the MC Method allows an additional
cross check on the Temperature Signal received from the vehicle.
Because the station can calculate the expected temperature from the
MC parameters and delivered enthalpy, it can cross reference this
with the temperature signal from the vehicle. The temperature
signal at the beginning of the fill procedure is generally constant
so by using the actual measured initial temperature and the
characteristic MC parameters, the vehicle fueling protocol can be
fully defined, and higher quality fill results can be achieved (as
reflected in both SOC and fill time).
[0150] The MC Method Fueling Protocol can be utilized
comprehensively by the station, for Identification Fueling,
Non-Communication Fueling and Full Communication Fueling, resulting
in a fill protocol that is better adapted to the current
capabilities of both the vehicle and the hydrogen filling station
capabilities and takes into account the current operating
environment to achieve higher quality fills.
[0151] An aspect of using the MC Method is the accurate prediction
of the mass average enthalpy that will be delivered to the tank
during a refueling procedure or event. As shown in FIG. 21, a 10K
error in the mass average temperature can result in a 6K to 8K
error in T.sub.final, so it is important to accurately predict the
enthalpy of the upcoming fill. In connection with the MC Method
testing, a Runge-Kutta approximation was developed for average
hydrogen enthalpy at the nozzle from stations using pre-cooling as
illustrated below in Equation [12].
h _ = 1 4 [ ( h ( T precooling , ( P StationInit ) + h ( T
precooling , ( P StationInit + P StationFinal - P StationInit 4 ) 2
) + ( h ( T precooling , ( P StationInit + P StationFinal - P
StationInit 4 ) + h ( T precooling , ( P StationInit + 2 P
StationFinal - P StationInit 4 ) 2 ) + ( h ( T precooling , ( P
StationInit + P StationFinal - P StationInit 4 ) + h ( T precooling
, ( P StationInit + 3 P StationFinal - P StationInit 4 ) 2 ) + ( h
( T precooling , ( P StationInit + 3 P StationFinal - P StationInit
4 ) + h ( T precooling , ( P StationInit + 4 P StationFinal - P
StationInit 4 ) 2 ) ] [ 12 ] ##EQU00011##
Where
[0152] T.sub.precooling=Expected Precooling Temperature
[0153]
P.sub.StationInit=P.sub.init+.DELTA.P.sub.StationInit=Initial
Hydrogen Tank Pressure+Initial Staiton Pressure Drop
[0154]
P.sub.StationFinal=P.sub.Final+.DELTA.P.sub.StationFinal=Final
Hydrogen Tank Pressure+Final Station Pressure Drop
[0155] During testing, it was found that .DELTA.P.sub.StationInit=5
MPa if the initial tank pressure was 2 MPa, 2 MPa if the initial
tank pressure was 17 MPa, and 1 MPa at higher initial pressures.
.DELTA.P.sub.StationFinal was assumed to be 1 MPa in all cases.
Therefore, the algorithm may be modified to reflect, more
accurately, the conditions and performance of a particular station.
In an illustrative example, the station builder or operator may
modify the algorithm to more accurately reflect the conditions and
performance of the station.
[0156] During several test fills, deviations were noted between the
pre-cooler output temperature and the actual temperature delivered
at the nozzle. These deviations tended to follow a relationship
with mass flow rate and pre-cooling level as illustrated in FIG.
23. In general, the higher flow rates and/or larger differences
between the pre-cooling and ambient temperatures will be reflected
in greater temperature deviations between the nozzle temperature
and the pre-cooler set temperature. Therefore, such factors can be
taken into account in the MC Method. In a non-limiting example,
each station builder or operator may determine this relationship(s)
for the range of expected operating conditions and parameters in
order to select an appropriate pre-cooling level that will
typically provide customer friendly refueling times. This
flexibility is one of the benefits of the MC Method--it allows the
station to calculate the appropriate fill time for a particular
pre-cooling temperature based on the conditions of that fill and
the capabilities of the tank itself.
[0157] The algorithms utilized in practicing the MC Method are
provided below. As the vehicle approaches the hydrogen filling
station, the vehicle provides the station, via RFID, IRDA or other
communication method, parametric data for a MC Method fill
procedure. The parametric data can include, for example: [0158] NWP
[0159] Tank Volume (or the station can calculate it with a pressure
pulse) [0160] Hot Soak Assumption [0161] Cold Soak Assumption
[0162] Constants of MC Equation [0163] Other parameters as desired
(Max Temp Allowed, Fastest Fill Rate Allowed, Max .rho..sub.Target
Allowed, etc.)
[0164] Even if none of the parameters are communicated, the station
can use the MC Method to conduct the fill by utilizing default
Constants of the MC Equation as derived from SAE TIR J2601, and the
default Hot Soak, Cold Soak assumptions of SAE TIR J2601.
[0165] Step 1--Calculate the Fueling Time Using Hot Soak
Assumption
T init = T ambient + .DELTA. T hot ##EQU00012## .DELTA. t = 0
##EQU00012.2## m init = V .times. .rho. initial ( T initial , P
initial ) ##EQU00012.3## m cv = V .times. .rho. target
##EQU00012.4## m add = m cv - m initial ##EQU00012.5## u initial =
u initial ( T initial , P initial ) ##EQU00012.6## h average = m
add h i ( T , P ) m add ##EQU00012.7## u adiabatic = m initial u
initial + m add h average m cv ##EQU00012.8## T adiabatic = T (
.rho. target , P adiabatic , u adiabatic ) ##EQU00012.9## M C = C +
A U adiabatic U init + g ( 1 - - k .DELTA. t ) j ##EQU00012.10## T
Final = m cv C v T Adiabatic + M C T Initial ( M C + m cv C v )
##EQU00012.11##
[0166] If T.sub.final>85C (or other user specific limit),
.DELTA.t=.DELTA.t+10 s.
[0167] Iterate from the top of Step 1.
[0168] As a practical measure, P.sub.adiabatic can be assumed to be
the MAWP with only a very small error, since internal energy has a
very weak relationship with pressure.
[0169] Step 2--Calculate the Pressure Target Using Cold Soak
Assumption
T init = T ambient - .DELTA. T cold ##EQU00013## m init = V .times.
.rho. initial ( T initial , P initial ) ##EQU00013.2## m add = m cv
- m initial ##EQU00013.3## u initial = u initial ( T initial , P
initial ) ##EQU00013.4## h average = m add h i ( T , P ) m add
##EQU00013.5## u adiabatic = m initial u initial + m add h average
m cv ##EQU00013.6## T adiabatic = T ( .rho. targer , P adiabatc , u
adiabatic ) ##EQU00013.7## M C = C + A U adiabatic U init + g ( 1 -
- k .DELTA. t ) j ##EQU00013.8## T Final = m cv C v T Adiabatic + M
C T Initial ( M C + m cv C v ) ##EQU00013.9## P Target = P ( .rho.
target , T Final ) and .rho. target = 100 % S O C ##EQU00013.10## C
P R R = P Target - P init 180 s + .DELTA. t ##EQU00013.11##
[0170] Step 3 (if necessary)--Calculate the Expected Result
T init = T ambient ##EQU00014## m init = V .times. .rho. initial (
T initial , P initial ) ##EQU00014.2## m cv = V .times. .rho.
target ##EQU00014.3## m add = m cv - m initial ##EQU00014.4## u
initial = u initial ( T initial , P initial ) ##EQU00014.5## h
average = m add h i ( T , P ) m add ##EQU00014.6## u adiabatic = m
initial u initial + m add h average m cv ##EQU00014.7## T adiabatic
= T ( .rho. target , P adiabatic , u adiabatic ) ##EQU00014.8## M C
= C + A U adiabatic U init + g ( 1 - - k .DELTA. t ) j
##EQU00014.9## T Final = m cv C v T Adiabatic + M C T Initial ( M C
+ m cv C v ) ##EQU00014.10## p Target = P ( .rho. target , T Final
) ##EQU00014.11## If
P.sub.Target>P.sub.TargetColdSoak,.rho..sub.Target=.rho..sub.Target-0.-
001 g/L.
[0171] Iterate From the Top of Step 3.
[0172] A hydrogen station can maintain a database of MC parameters
that have been communicated to the station, and use the lowest
performing MC parameter, tank volume, and lowest initial SOC %
historically observed, to set the pre-cooling temperature for the
system in order to achieve a fast fueling rate given ambient
temperature. In this way a station can keep the pre-cooling
temperature set at an economically optimal level.
[0173] Although the MC Method was developed and has been described
with an emphasis on filling vehicle hydrogen tanks at hydrogen
filling stations, modification of the MC Method to improve its
performance in connection with fueling hydrogen busses or fueling
systems with cryogenic gasses or liquids is certainly contemplated.
Similarly, it is anticipated that the basic MC Method could readily
be adapted for use in conjunction with compressed natural gas
vehicle fueling, or fast filling of vessels involving any
industrial gas and/or for calculating the resulting temperature of
any process in which a pressurized gas is injected into a pressure
vessel. The applicability of the MC Method and the associated
constants reflecting the thermodynamic properties and behavior for
other processes can be determined by applying a similar test matrix
as set out above in connection with compressed hydrogen tank
refueling for automobiles.
* * * * *